Start With the Base Rate, Always
The single highest-return forecasting habit is asking how often this kind of thing usually happens before thinking about this particular case.
If you only adopt one habit from this category, adopt this one. Before forming any view about a specific question, ask how often things of this kind happen in general.
That number is the base rate, and starting from it is the difference between forecasting and storytelling.
Why it works
Specific cases come with vivid detail. The detail feels informative and mostly is not. A question about whether a particular company ships a particular product by a particular date arrives with a rich narrative, and the narrative crowds out the far more useful fact that announced timelines of this type slip most of the time.
The base rate is the outside view. It asks what happened the last hundred times something looked like this. It is unglamorous, it ignores everything interesting about the case in front of you, and it beats most careful case-specific reasoning.
Getting one
Define the reference class. What counts as "this kind of thing". Too broad and it is uninformative; too narrow and there are three examples. This judgement is most of the skill.
Count. How often did the thing happen in that class. Often this is a quick search and an approximate number, which is fine — the difference between 20% and 25% matters far less than the difference between "rarely" and "usually".
Write it down before looking at the market. Otherwise you will anchor on the price and produce a number that is the price with a small adjustment, then congratulate yourself on agreeing with it. Keeping a forecast journal is the discipline that enforces this.
Then adjust, carefully
The base rate is the starting point, not the answer. Adjust for what is genuinely different about this case, and adjust less than you want to.
Adjust for structural differences. If the reference class is "announced timelines" and this one has already been delayed twice, that is a real signal.
Do not adjust for narrative. Confidence in a press release is not evidence. Neither is how compelling the argument you just read was.
Bound the adjustment. If your base rate is 20% and you find yourself at 75%, ask what you know that the reference class does not. Occasionally the answer is good. Usually it is that the story was vivid.
Base rates and market prices
Once you have a base rate, the market price becomes readable. If the base rate is 15% and the market is at 45%, one of three things is true: the market knows something you do not, your reference class is wrong, or the market is wrong. In that order of likelihood.
That ordering is the useful part. The first response to disagreeing with a market should be to look for what you are missing, not to trade. The limits of crowd wisdom covers when the third option becomes plausible.
Where markets fail base rates
Markets are populated by people, and people neglect base rates too. The documented places this shows up:
Novel-seeming events. Anything without an obvious reference class gets priced on narrative.
Long tails. Low-probability outcomes are systematically overpriced, partly because the base rate for a vivid unlikely event feels higher than it is. The favourite-longshot bias.
Recency. A reference class that includes one dramatic recent example gets overweighted.
The practice
For the next ten markets you look at, write down a base rate before you look at the price. Then look. Keep a record of both.
After thirty of those you will have something genuinely useful: evidence about whether your reference classes are any good. That is the beginning of calibration, and it is the only way to find out whether you are forecasting or narrating.
Next
Calibration curves — how to check whether your numbers mean anything.
Educational content only. Nothing here is financial advice, a solicitation, or an offer to trade. Prediction markets can move against you and a position can lose its entire value. Where this guide describes a difference between venues it describes a theoretical cross-market edge in observed prices, which is not a profit: fees, spreads, capital lockup, transfer time and differing settlement rules all come out of it first.
Keep reading
Forecasting and Calibration · 3 min readCalibration Curves and How to Read Your OwnA calibrated forecaster is right 70% of the time when they say 70%. Building the curve, reading its shape, and what each kind of deviation tells you.
Forecasting and Calibration · 3 min readKeeping a Forecast Journal That Is Actually UsefulMost forecasting improvement comes from records, not from cleverness. What to write down, what to leave out, and how to review it without flattering yourself.
Forecasting and Calibration · 3 min readWhere the Wisdom of Crowds Runs OutAggregated opinion beats most individuals most of the time. The conditions it depends on are specific, and prediction markets sometimes fail to meet them.
Kalshi · 3 min readHow Kalshi Fees Work, and Where They Hurt MostKalshi's trading fee is not a flat percentage. It is shaped like a curve that peaks at fifty cents, and that shape changes which trades are worth doing.